Abstract Purpose This scoping review aimed to systematically map and critically describe the current evidence on the role of the gut microbiome as a biomarker in oncology, including microbiome-based predictive models and microbial signatures associated with treatment response, toxicity and disease course, and to identify methodological gaps and challenges for clinical translation in precision medicine. Methods This scoping review was conducted following the Joanna Briggs Institute (JBI) guidelines for scoping reviews and reported according to the PRISMA Extension for Scoping Reviews (PRISMA-ScR). The search was performed in five electronic databases (PubMed/MEDLINE, Web of Science, Scopus, SciELO, and LILACS) using a structured PICO strategy. Studies involving adult cancer patients undergoing systemic oncological therapies (including chemotherapy, immunotherapy and combined regimens), with gut microbiome analysis and the investigation, development or validation of microbiome-based biomarkers or predictive models, were included. Results The literature demonstrates that specific microbial taxa significantly influence the efficacy of immunotherapies (e.g., AUCs up to 0.88 for ICI response prediction) and chemotherapies, and modulate toxicity (e.g., mucositis reduction from 47.1% to 25% with probiotics). Microbiome-based predictive models often outperform clinical markers (e.g., AUC of 0.88 vs. 0.50 in urothelial carcinoma), and variations in microbiota composition can predict disease progression. The literature mapping of the 20 included studies demonstrates that specific microbial taxa significantly influence the efficacy of immunotherapies (e.g., AUCs up to 0.88 for ICI response prediction) and chemotherapies. Regarding toxicity, while the review focuses on baseline biomarkers, exploratory intervention-based data were addressed, showing that a probiotic cocktail reduced Grade 3–4 oral mucositis from 47.1% to 25%. Furthermore, microbiome-based predictive models demonstrated enhanced discriminatory accuracy in predicting patient outcomes compared to standard clinical classification markers alone (e.g., achieving an Area Under the Curve AUC of 0.88 vs. 0.50 for clinical factors in urothelial carcinoma), though these models remain in early exploratory stages. Conclusion Overall, the evidence suggests a growing interest and potential for microbiome-based predictive models in oncology; however, their clinical translation remains limited by methodological heterogeneity, insufficient external validation, and incomplete mechanistic understanding.
Menezes et al. (Sat,) studied this question.